There is a long-standing literature showing the presence of conditioned opiate-like and opiate-withdrawal responses in opiate dependent animals and humans. These data suggest that learned responses may be particularly important to relapse and readdiction yet no current treatment method directly addresses this aspect of addiction. Previous work has shown that attempts to abruptly extinguish conditioned responses in addicts lead to a high drop-out rate. We now have pilot data indicating that a proceudre including gradual extinction of conditioned responses to drug-related stimuli integrated with cognitive therapy is well-accepted by methadone treatment patients. In order to test whether extinction of conditioned responses has usefulness in a treatment program, we propose a randomized clinical trial among 120 stabilized methodone patients who desire impatient detoxification. Three groups will be studied. The cognitive/extinction group (CE) will receive a hierarchical presentation of stimuli involved with drug procuring and self-administration. The conditioned responses evoked by these stimuli will be reduced by repetition in the absence of drug effects and by relaxation training. This will be integrated with cognitive therapy. The control groups will spend equivalent time receiving either cognitive therapy with relaxation (CT) or drug counseling only (DC). A full range of pre-, during-, and post-treatment as well as follow-up measures will be used. The purpose of this study will be to determine 1) whether conditioned responses in opiate addicts vary over time and whether such responses relate to clinical outcome; 2) whether a gradual extinction program can reduce or eliminate these conditioned responses; 3) whether there is any benefit to adding treatment of conditioned responses to standard treatment and, if so, which patients will benefit.